Notice: Trying to get property 'display_name' of non-object in /var/www/html/wp-content/plugins/wordpress-seo/src/generators/schema/article.php on line 52
keyboard_arrow_uptop

After taking a detailed look at the Dodgers’ acquisition of Derek Lowe, a number of readers wrote in to discuss the hundreds of new field-level seats being added to Dodger Stadium. The general question was: “Did I consider this in my research, and if not, how do I think it will affect Lowe and the Dodgers in general?” The answers I provided over e-mail were: “No, I didn’t, and it should increase general offense in the park. Because that should mainly be a result of a few more foul pop-ups lost to the seats, Lowe would be affected less by these changes than other pitchers because he’s such an extreme groundball pitcher.”

This got me wondering about parks and the application of park factors. Park factors are not simply for adjusting runs; they can also be applied to individual statistics such as singles, doubles, triples, home runs, etc. While most parks see a consistent difference in most offensive categories (Coors Field and Bank One Ballpark increase all offensive measures while decreasing strikeouts), some stadiums have differing effects on similar parts of offense. As I discussed last time, Dodger Stadium differs from Fenway Park in that it suppresses doubles and triples, but allows more home runs. Additionally, Comerica Park and SBC Park allow nearly double the league average of triples while suppressing all other forms of offense.

Generally what happens at this point is people look at those factors and say “Well, that makes sense. Fenway Park’s Green Monster and cavernous center field should increase doubles and triples while Dodger Stadium’s shorter center field allows the center fielder to play more shallow and cut off a few more drives to the power alleys.” Everyone nods in agreement, noting how that makes sense.

Of course, the one thing people rarely talk about with park factors is the infield, because it’s the only part of the ballpark mandated to be exactly the same dimensions everywhere. Outfields can be all sorts of sizes and shapes within the league guidelines (and sometimes not), but it’s always 90′ to first, 60’6″ to the mound, and, if my math is correct, 127’3″ from home to second. While the surface can be different, and while some teams intentionally keep their grass short or long, infields are all very similar and generally considered not to have as much of an effect on park factors as outfield dimensions.

It stands to reason then that groundball pitchers should be less affected by park factors than flyball pitchers. By keeping the ball on the ground, groundball pitchers should see significantly more balls fielded in the more regulated infield and fewer balls rattling around the unique confines of each outfield.

While park factors for left- and right-handed players have been used in the past, groundball/flyball ratio is not the same binary system that lefty/righty is. There’s no clean cutoff for differentiating groundball pitchers from flyball pitchers. Simply using players above or below league average every year likely would not yield enlightening
results because groundball/flyball rates form a nice bell curve with most of the league within a very small range around the mean. Putting players with a GB/FB ratio .01 higher than the mean on one side and those .01 lower on the other would give us two groups of very similar pitchers.

Instead, looking at only players on the extremes of the groundball/flyball scale will give us a better idea of whether or not groundball pitchers are less affected by park factors than flyball pitchers. Thus, for the past four seasons, I’ve compared the home and road stats of all pitchers who threw at least 20 innings both at home and on the road. From that group, I’ve then broken them into three groups for each season: groundball pitchers, neutral pitchers and flyball pitchers. These groups have been defined by being one standard deviation either above or below the mean for each season.

For example, of the pitchers who met the innings qualifications in 2004, the mean GB/FB ratio was 1.30 with a standard deviation of .55. Thus, all pitchers with a GB/FB ratio above 1.85 are part of the groundball group while all pitchers below .75 are part of the flyball group. Everyone in between those numbers is in the neutral group. Generally, one standard deviation above and below the mean will encapsulate about 65% of the sample, so each extreme group has only about 17-18% of the total sample. There’s some concern about small sample size, but I’m more concerned with each group having too many pitchers with more neutral splits.

With the groups established, we now need to establish how to determine if the groundball group is less affected by park factors than the flyball group. To answer this question, I’ve computed each pitcher’s individual park factors, comparing their performance at home versus their performance on the road, weighted for playing time and centering the park factors around 1.00. Thus, if a pitcher were completely immune to park factors, he would have a 1.00 park factor for every statistic examined. Obviously no pitcher fits that bill, but if groundball pitchers are less affected by park factors than flyball pitchers, their park factors should be closer to 1.00 than those of flyball pitchers.

There’s one additional small issue with this method. If the sample of pitchers happens to be from a more extreme set of parks than the other groups, the variance found between the pitchers’ performances might be the result of a more extreme park rather than the groundball/flyball tendencies of the group. For instance, Coors Field had a runs park factor of 1.31 in 2004 (increasing run scoring by 31%) while Turner Field was exactly neutral at 1.00. Two sample pitchers, one a groundball pitcher and one a flyball pitcher and each with runs park factors of 1.10, would appear to support the idea that both groundball and flyball pitchers are equally affected by park factors. However, the pitcher in Coors has been affected significantly less than the pitcher in Turner Field.

To correct for this, each pitcher’s individual park factor will be compared to their composite park factor for that season. Comparing on a ratio basis would not only skew results, but also yield division by zero, so differences in actual performance and park factors will be compared on an absolute basis. Taking our above pitchers, the pitcher in Coors Field would have a value of -0.21 whereas the Atlanta pitcher would have a value of 0.10. Negative values indicate the pitcher was less affected by the park than expected.

Here’s how each group averaged in each of the last four seasons for a variety of park factors:


Group Year    H   ER    R   BB   HR   SO
GB    2001  .23  .57  .57  .32  .41  .11
GB    2002  .14  .29  .22  .19  .61  .19
GB    2003  .13  .25  .27  .35  .55  .17
GB    2004  .21  .51  .47  .25  .34  .22

FB    2001  .19  .30  .31  .40  .31  .11
FB    2002  .22  .56  .53  .31  .82  .13
FB    2003  .17  .75  .79  .39  .48  .16
FB    2004  .18  .54  .44  .31  .49  .13

Diff  2001  .04  .27  .26 -.08  .10  .00
Diff  2002 -.08 -.27 -.31 -.12 -.21  .06
Diff  2003 -.04 -.50 -.52 -.04  .07  .01
Diff  2004  .02 -.03  .03 -.05 -.15  .09

Each number in the chart is the average of how much more variable each group was than the league-wide park factor that season. Because individual pitchers are much more variable than a three-year park factor, all of the numbers for the two groups are positive. In the third section (marked by “Diff”), the two groups are compared, subtracting the flyball group from the groundball group. If the groundball group were less variable than the flyball group, the numbers in this section would be negative.

One look at those final four rows doesn’t yield the results I was expecting to find. The groundball group is less variable than the flyball group in 13 of the 24 samples, barely over half. Because our sample of extreme groundball pitchers appears to show as much variance in their performances at home and on the road as the extreme flyball pitchers, we cannot conclude from this analysis that groundball pitchers are less subject to the adjustments of park factors than flyball pitchers. Even when looking at categories like doubles and triples, the groundball group doesn’t show significantly less variance than the flyball group.

I must admit that I’m a little surprised that this appears to be the case. Intuitively, it seems that keeping the ball on the ground and in the infield more often should leave pitchers less exposed to the unique features of each park’s outfield. With GB/FB ratio appearing to have no effect on park factors, we must concede that the other unique aspects of each park–weather, hitting background, surface, altitude, etc.–cause similar amounts of variance for both types of pitcher.

Getting back to Lowe, while I had been e-mailing that I didn’t believe the extra seats in Dodger Stadium’s foul ground would affect him as much as flyball pitchers, I can’t find any information in this analysis to support that claim. Instead, it appears that those seats should affect Lowe as much as any other pitcher in Chavez Ravine, despite his noted extreme groundball tendencies. It appears that Lowe will be able to take advantage of the rest of the offensive-destroying influences of Dodger Stadium. That still does little to justify his contract.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe